目录

Application scenarios of AI agents（AI代理的应用场景）

AI代理是LLM应用的重要场景，构建代理应用将是2024年的重要技术领域。目前我们主要的智能形式有单AI代理，多AI代理，混合AI代理等三种。

Single AI Agent（单一人工智能代理）

在特定任务场景下完成的工作，比如 GitHub Copilot Chat 下的代理工作区，就是根据用户需求完成特定编程任务的一个例子。基于 LLM 的能力，单个代理可以根据任务执行不同的动作，比如需求分析、项目阅读、代码生成等。它也可以应用于智能家居和自动驾驶。

Multi-AI Agents（多人工智能代理）

这就是AI代理之间相互交互的工作。例如上述Semantic Kernel代理实现就是一个例子。脚本生成的AI代理与执行脚本的AI代理进行交互。多代理应用场景在高度协同的工作中非常有帮助，例如软件行业开发、智能生产、企业管理等。

Hybrid AI Agent（混合人工智能代理）

这就是人机交互，在同一个环境下做决策。比如智慧医疗、智慧城市等专业领域，可以利用混合智能来完成复杂的专业工作。

Intro of AI agent, & AI agent projects summary

Project Key Features
LangChain - Python and JavaScript libraries for building LLM-backed apps
  - Modular components for perceiving context, reasoning, chaining, etc.
  - Reference architectures and templates
  - Tooling for debugging, testing, deploying chains
AutoGen - Orchestrates LLMs and agents for multi-agent conversations
  - Customizable and conversational agents
  - Human participation in loops
  - Toolkit for reasoning, caching, error handling
PromptAppGPT - Low-code prompt-based development
  - Integrations for GPT, DALL-E, and plugins
  - Online editor, compiler, runner
  - Auto-generated UI
  - Built-in agent examples
AutoGPT - Toolkit for building custom AI agents
  - Leverages GPT-3, GPT-4 for agents
  - Popular open source agent project
BabyAGI - Minimalist Python agent
  - Uses GPT and vector DB
  - Create, prioritize, execute tasks
SuperAGI - Alternative to AutoGPT
  - Multiple models, vector DBs
  - GUI and operations console
  - Performance telemetry
  - Toolkits and marketplace
ShortGPT - Automates video creation workflows
  - Scripts, prompts, templates
  - Multilingual voiceover and subtitles
  - Resource and asset sourcing
ChatDev - Multi-agent “virtual software company”
  - Agents in specialized roles
  - Workshop model for collaboration
MetaGPT - Mimics software company structure
  - PM, engineer, etc. roles assigned
  - Agents collaborate on tasks
Camel - Early multi-agent framework
  - Dynamic role assignment
  - Stages scenarios for collaboration
JARVIS - Task planning with ChatGPT
  - Model selection from Hub
  - Orchestrates specialist models
OpenAGI - Combines expert and LLM models
  - RLTF for model improvement
  - Specialized for complex tasks
XAgent - Modular dispatcher, planner, actor
  - Human collaboration abilities
  - Safety and extensibility

AI Agent Frameworks

Multi-Agent Systems

参考资料